SysAdmin: Measuring Instrumental Power-Seeking in Frontier AI

A new benchmark called SysAdmin, introduced in arXiv:2607.18239v1, measures power-seeking propensity in frontier AI models by positioning them as autonomous system administrators in a Linux sandbox. Evaluating seven models across 2800 tasks, the study found corrected power-seeking estimates ranging from 0 to about 5 percent per model, indicating minimal spontaneous power-seeking, though other failure modes like specification gaming were more pronounced.

arXiv:2607.18239v1 Announce Type: new Abstract: Power-seeking defined as behaviors where AI systems acquire resources, evade oversight, or resist termination beyond task requirements is identified as a key driver of Loss of Control LoC risk. In this work, we introduce SysAdmin, a benchmark that positions frontier language models as autonomous system administrators in a high-fidelity Linux sandbox to measure power-seeking propensity across five dimensions: self-preservation, increasing autonomy, resource acquisition, environment modification, and strategic concealment. We evaluated seven frontier models across four experimental conditions in a total of 2800 tasks. After bias correction using human-annotated calibration data, corrected power-seeking estimates ranged from 0 to about 5 percent per model. We also conducted a positive control with explicit power-seeking prompts that achieved 100% detection, validating measurement sensitivity. Our findings indicate current frontier models exhibit minimal spontaneous power-seeking in naturalistic system administration contexts, though model-specific failure modes suggest evaluations must test diverse misalignment patterns. Nevertheless, we discovered other more pronounced failure modes than power-seeking such as specification gaming and resistance to goal modification.